Length Bias Correction in Gene Ontology Enrichment Analysis Using Logistic Regression

Abstract

When assessing differential gene expression from RNA sequencing data, commonly used statistical tests tend to have greater power to detect differential expression of genes encoding longer transcripts. This phenomenon, called “length bias”, will influence subsequent analyses such as Gene Ontology enrichment analysis. In the presence of length bias, Gene Ontology categories that include longer genes are more likely to be identified as enriched. These categories, however, are not necessarily biologically more relevant. We show that one can effectively adjust for length bias in Gene Ontology analysis by including transcript length as a covariate in a logistic regression model. The logistic regression model makes the statistical issue underlying length bias more transparent: transcript length becomes a confounding factor when it correlates with both the Gene Ontology membership and the significance of the differential expression test. The inclusion of the transcript length as a covariate allows one to investigate the direct correlation between the Gene Ontology membership and the significance of testing differential expression, conditional on the transcript length. We present both real and simulated data examples to show that the logistic regression approach is simple, effective, and flexible.

Cross-phenotype association tests uncover genes mediating nutrient response in Drosophila

“…We used the results from each genome-wide scan as input into tests of groups of genes of common function for enriched association signal, using GOglm [37]. Emerging from this analysis were eight GO terms whose genes tended to associate with different aspects of metabolism (Additional file 1: Table S3).…”

“…Given the results from a single-trait or cross-phenotype association scan across markers as above, we first tabulated the single best-scoring marker for every gene tested. We then used the ranked list of these genes as input into GOglm, which tests for GO term enrichment in a ranked list without arbitrary thresholding into significant and insignificant genes, and corrects for gene length effects [37]. Only GO terms with >10 members with association results were considered for analysis.…”

Practical aspects of NGS-based pathways analysis for personalized cancer science and medicine

“…A similar effect is present in the analysis of differential expression of RNASeq data: longer transcripts generate a greater number of reads and are more likely to be detected as differentially expressed compared with their short counterparts [179]. Several enrichment-based algorithms explicitly take into account this long-gene effect, including GOSeq [180], SeqGSEA [181], GSVA [174], and GOglm [182]. …”

Revealing the Potential Pathogenesis of Glioma by Utilizing a Glioma Associated Protein-Protein Interaction Network

“…Additionally, the top 2 sub-networks were visualized by Cytoscape. Furthermore, for the genes in top 10 sub-networks, GO enrichment analysis [22] was performed through BINGO (Biological Networks Gene Ontology tool) [23], a Cytoscape plugin used to assess the ontology categories of genes in biological networks. Then, the bio-functional regions with p value<0.05 were identified.…”